clear 
use "/Users/marclerchenmueller/Dropbox/BMJ_Gender_Promotion/BMJ_Dataverse/HDV_data/pubmed_articles_90pct_genderdesig.dta" 

***************POSITIVE FRAMING ANALYSES*******************************************************************************************************************************************************************

***Linear probability models for life science research, overall and stratified by journal impact factor category 

areg i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017, absorb(issn)
margins i.l_female
margins r.l_female

areg i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if (jif > 10), absorb(issn)
margins i.l_female
margins r.l_female
**FIGURE 4

areg i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if (jif <= 10), absorb(issn)
margins i.l_female
margins r.l_female
**FIGURE 4

***************CITATION ANALYSES************************************************************************************************************************************************************************

**Note: When the dependent variable is log transformed, the coefficient on the dummy variable for positive framing is to be exponentiated to get to %change in citations as presented in article

areg log_cites i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if (jif > 10), absorb(issn) 
estimates store pubmed_high_jif
estimates table pubmed_high_jif, eform  
**FIGURE 5

areg log_cites i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if (jif <= 10), absorb(issn) 
estimates store pubmed_low_jif
estimates table pubmed_low_jif, eform  
**FIGURE 5

*Sensitivity analysis: Poisson regressions 

poisson cites i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if jif > 10, irr vce(robust)

poisson cites i_positive i.l_female pct_female n_authors med_mesh_pctile /*
*/ fe_2003 fe_2004 fe_2005 fe_2006 fe_2007 fe_2008 fe_2009 fe_2010 fe_2011 fe_2012 fe_2013 fe_2014 fe_2015 fe_2016 fe_2017 if jif <= 10, irr vce(robust)

